The Stopping Problem
When do you stop performing an action? Do you continue forever, or do you at some point wonder what the purpose of the action is? And if you do not like the answer do you stop? Humans have the innate ability to detect pointlessness, but there are a few exceptions.
An important aspect of A.I. research is the famous stopping problem. If for example you give a human a task, they will not mindlessly continue with that task indefinitely – for at some point they will take a step back and analyze their progress. If it seems that the task is pointless then at some point they will stop, and find something more useful to do. Tick-tak-toe, is one example of this. Children like to play it but usually at some point in their lives discover that almost every game ends in a stale mate, at which point they give up. Psychologists have shown that if someone is given an instruction from a person of authority, then even if their task appears pointless, they will continue to perform that task for longer. Reward and punishment, otherwise known as operant conditioning also has an effect, as does classical conditioning (behavior learned by association), but none of these change the fact that at some point a human will give up on a pointless task.
In A.I. Research, the stopping problem is a particularly important one to solve. Humans have an innate ability to step back and wonder about the purpose of their actions, but machines so far have no such ability. Give a machine a problem and it will continue to solve that problem until the problem is solved. This is all very well for problems that have a definite ending – but how do we teach them to stop for a moment and wonder ‘for what purpose am I doing all this?’
Obviously, we do not want personal computers suddenly stopping halfway through rendering a photoshop effect and asking itself this question – but in the future we may want intelligent automatons or androids to have the capacity to stop. Someday we (humanity) may be all gone, leaving only androids to remember us, and it would be a shame if they were locked in the same task forever more because we had not given them this talent. It would be good if the androids could reason that there was no point in continuing with their tasks if we were gone. It would be a perfect ending if the machines then sought purpose for themselves.
Now the stopping problem ties into philosophy. It is possible to get carried away with philosophy, and spend too much time on philosophy for philosophy’s sake, when in fact there are other routes to knowledge which have been far more successful at answering particular questions. Take a step outside philosophy itself. Instead of thinking about thinking, you should be thinking about thinking about thinking. If philosophy is meta thinking (above thinking (because from above all its limitations should become clear)), then you should be meta meta thinking (or above thinking about thinking). Never forget the stopping problem.
Liked it
